Sustainably sustaining (online) fashion consumption: Using influencers to promote sustainable (un)planned behaviour in Europe ’ s millennials

This study explores if, how and through what channels millennials ’ sustainability values translate into action when it comes to fashion garments. By testing a research model on 448 European millennials, it contrasts extant theories of planned behaviour, finding that purchase intent is often guided by unintentional, non-linear processes wherein trust in intermediaries such as celebrity influencers, rather than the fashion retailers ’ corporate social responsibility (CSR) approach condition behaviour. For fashion retailers, it suggests the strategic use of influ-encers to (un)consciously market sustainable garments. Its novelty is built on sustainably sustaining fashion consumption in a post-pandemic world, characterised by increased online sales.


Introduction
Fashion can be viewed as the artistic means for consumers to express themselves by following the latest trends or adopting niche styles.The fashion industry is not only growing, given the rise of disposable income and/or credit institutions in many parts of the world (see Cabigiosu, 2020), it is also under increasing stakeholder pressure to become more sustainable, given the multiple social and environmental challenges associated with dominant fast fashion business models and unsustainable consumption patterns (see Pedersen and Andersen, 2015).Nevertheless, more research is needed on sustainable knowledge for the fashion industry for and by multiple actors (see Kong et al., 2016).This is arguably because most extant research focuses on consumer perspectives (e.g.McNeill and Venter, 2019;Youn et al., 2021), rather than on the relationship between consumers, retailers and other intermediaries in the online retail context (see Johnstone and Lindh, 2021); and particularly on what fashion retailers can do to promote (or sustain) sustainable purchase behaviours (see Yang et al., 2017;Lee et al., 2020).
Recent corporate social responsibility (CSR) scandals within the fashion industry have put pressure on fashion producers and retailers by more socially and environmentally aware consumers (see Henninger et al., 2016).Millennials (i.e.those born between the early 1980s and the mid-to-late 1990s [see Bucic et al., 2012;Mangold and Smith, 2012]) are proposed as one generational cohort of consumers that increasingly values socially responsible behaviour (see e.g.Heo and Muralidharan, 2019;Lu et al., 2013;Wang et al., 2018).Concurrently, millennials are generally considered the biggest (Debevec et al., 2013), most impressionable (Pomarici and Vecchio, 2014) and fashion-conscious consumer group (Samala and Katkam, 2019), thus making them a key market for fashion retailers (Gerardo, 2018) given that much of their disposable income is spent on fashion (see Samala and Singh, 2019).Together, this has led to an increase in fast fashion garments that have serious social and environmental impacts due to the associated rapid rates of rapid and unsustainable production (see Mukendi et al., 2020).
Extant research paints an unclear picture of the translation of sustainability values into millennials' purchase behaviour.Various studies have found that while millennials are positive towards sustainability, this attitude does not translate into action (e.g.Bernardes et al., 2018).More specifically, regarding fashion, various studies note an ethical paradox for millennials between their increasing sustainability values and beliefs, and their fashion needs or wants in terms of fashion consciousness (see e.g.Pencarelli et al., 2020;Rolling and Sadachar, 2018).Contrasting the assumption that millennials are a socially responsible generation and building on the intention-behaviour gap, Vuong and Nguyen (2018) indicate that for fashion garments, self-image often takes precedence over sustainability concerns in the purchase decision-making process (see also Iran et al., 2019).This suggests that the 'sustainable' purchase decision-making process for millennials may not entirely conform to linear or rational models of planned behaviour and that other factors may explain behavioural pathways.For example, beyond the fashion sector, Johnstone and Lindh (2018) find that millennials are unconsciously influenced by peers in the online retail context and that marketing products as sustainable per se is not conductive to increased sales.Rather, they suggest that millennials -as a generational cohort raised on technology (Eastman et al., 2014) -are affected by social pressure and intermediaries (e.g.online influencers) which incite unplanned (or unconscious) behaviour in terms of purchase intent (see also Johnstone and Lindh, 2021).Understanding how millennials' fashion purchase intent is affected through online communication channels is therefore important (Wolny and Mueller, 2013) given that 'fashion' spreads primarily through networking effects (Easley and Kleinberg, 2010), yet studies in the area are limited from an online retail stance.
Given that social media constitutes an important tool for retailers to reach consumers (Cluley, 2019;Krishen et al., 2016;Johnstone and Lindh, 2021) and studies into the use and role of influencers for informing millennials purchase intent in the online retail context are emerging (e.g., Chopra et al., 2020;Xu and Pratt, 2018), this study aims to better understand how sustainable business models can be achieved and what this means for extant marketing theories in an increasingly online retail context.This is furthermore important given that the global COVID-19 pandemic has pushed many fashion consumers online (see Koch et al., 2020), as well as the mounting pressure on the fashion industry to have more sustainable business models (see Todeschini et al., 2017;Brydges et al., 2020).Thus, this study asks if, how and through what channels millennials' sustainability values translate into action when it comes to fashion garments.For fashion retailers, such questions are necessary to help balance the perceived market needs and broader sustainability values of millennials, whilst continuing to make profit (see also Rahman et al., 2020).Reconciling the tension between the hedonistic needs and sustainability concerns of millennials furthermore constitutes an important question not only for research and practice, but also long-term sustainable futures in an increasingly digital world.
A research model is developed which suggests fashion consciousness, CSR and influencer influence affectand/or interact to affectmillennials' fashion purchase intent.This is tested through a two-part analysis to develop the constructs as well as the relationships between them.Importantly, the design of this model recognises that while millennials' CSR values may not directly translate into action, the fashion industry can utilise social media influencers to (un)consciously promote sustainable consumption.This contributes to a better understanding of how retailers can sustain the fashion industry, but in a socially responsive way through.Consequently, it places responsibility on fashion producers and retailers for a more sustainable industry in a postpandemic world, which will arguably be characterised by increasing online sales as an aftereffect.
Theoretically, the paradox between the millennials' axiological concerns about sustainability and the hedonistic factors associated with fashion consumption are explored by framing this research in connection with Johnstone and Lindh's (2018) theory of (un)planned behaviour.Notably, this theory contrasts extant consumer behaviour models by placing the emphasis on the retailers' ability to influencer consumer behaviour by (un)consciously building trust in online intermediaries such as influencers.This is novel as it moves beyond a production and/or consumer focus of said 'tensions' (i.e.reconciling increasing sustainability values with fashion consciousness as a personal decision-making process) to rather on how fashion retailers can (un)consciously promote sustainable fashion consumption within millennials; thus contributing to the development of theories of (un)planned behaviour in the online retail context.It furthermore implies that consumer behaviour may not be linear or rational as previously implied.

Building the research model
This study is framed around how the sustainability-fashion paradox regards a value-action gap when it comes to sustainable fashion purchases for millennial consumers.Building on previous studies, it appears that the intention to purchase sustainably and/or ethicallyis not necessarily reflected in the actual purchasing behaviour of consumers (see e.g.Bernardes et al., 2018;Han et al., 2017;Kumar et al., 2017).Nevertheless, going one step beyond this gap in terms of theoretical framing for the online retail context requires an acknowledgement of the other (f)actors at play, wherein the online decision-making process is affected by several exchange relationships, not only between the buyer and seller (Johnstone and Lindh, 2021).Therefore, this chapter first presents the theoretical framework which loosely guides the study, before building the hypotheses for the research model.Jones et al. (2000) defined online retailing as a process whereby actors interact using digital technologies or telecommunications to exchange value.More recent studies highlight that this exchange process is not necessarily unilateral or bilateral between buyers and sellers, but increasingly multilateral and multi-(f)actors, wherein consumer trust, decision-making and opinion are conditional on third parties in the form of, for example, review websites, influencers, online peers, social media posts, blogs, vlogs etc. (Johnstone and Lindh, 2021).Building on such assumptions, Johnstone and Lindh's (2018) theory of (un)planned behaviour is useful to help frame and explain (inter)actions -or exchanges -in the online retail context (see also Chen et al., 2015a) which is arguably distinct from the offline context and how extant models of planned behaviour present consumer behaviour.This helps contextualise the development of the research model by recognising the unique characteristics of online retail in terms of how interactions affect consumer trust and ultimately, their behaviour.

Theoretical framework
In brief, Johnstone and Lindh propose that -in contrast to Ajzen's (1991) theory of planned behaviour (TPB) -the causal relationship between intention and sustainable behaviour for millennial consumers in the online retail context is not linear (see Table 1 below).Rather, millennials are more sensitive to peer pressure and social processes which can affect initial intentions (see also Alexander and Sysko, 2013).Johnstone and Lindh (2018) build on previous work which suggests that increasingly, social or ethical purchases are guided by random or unconscious processes that cannot be fully explained by traditional theories of planned behaviour (see also Burke et al., 2014;Davies and Gutsche, 2016).They argue that the online retail context obscures the traditional, direct buyer-seller relationship and that for millennials, it is often (un)planned behaviourunconsciously promoted by online intermediaries or social interactionsthat guides purchase intent.Thus, they propose that the (over)reliance on deterministic models such as the TPB needs reconsideration for today's online retail context and the millennial generational cohort (see Table 1).
Furthermore, trust is an additional concept which holds interesting pathways for the online retail context given the value attributed to intermediaries such as social media influencers for millennials.Regarding e-commerce, Chen et al. (2015b, p. 289) define trust as "the subjective belief that the trusting party believes that the trusted party will behave by exhibiting integrity, ability, and benevolence".Beyond being built directly between the consumer and retailer's presence on social media, which may involve strategies such as web-shop layout, customer feedback forums and third-party certification, among others (Irshad et al., 2020), trust is also increasingly built through indirect pathways online.Johnstone and Lindh (2021) assert that behaviour or action in terms of purchase intent in the online retail context is positively affected by reviews/reviewers that the consumer perceives as trustworthy.Johnstone and Lindh (2018) further propose that the belief in influencers, is the mechanism of trust (see also Kim and Kim, 2021), thus making it integral to (un)planned behaviour in terms of purchase intent.Here, trust in the online retail context is built on cognitive processes between the retailer and the consumer wherein trust in the former by the latter is conditional on a third party (an individual (i.e.actor) or online communication channel (i.e.factor)) which guides behaviour (see Chen et al., 2015a).Thus, trust is proposed in relation to the level, intensity, number of social exchanges and value attributed to increasingly diverse (f)actors for online retail and serves an additional explanatory role in the conceptualisation and discussion of the research model proposed in this study.

Hypotheses' development
By keeping the concept of trust and theory of (un)planned behaviour in mind for the online retail context, the following section develops the hypotheses for the research model in relation to the sustainable-fashion consumption paradox for millennial consumers.

Fashion consciousness
The global fashion industry is under increasing stakeholder pressure to perform sustainably.It is also an industry that almost everyone can relate to given that -as consumers -we all make decisions on our clothing as one element of 'fashion'.Further still, sustainably sustaining the fashion industry is imperative to a better future for all in terms of the international sustainable development goals.Rahman et al. (2020) propose fashion as "an activity whereby an individual expresses personal tastes and interests through clothing".Arguably then, being 'fashionable' is inherently tied to the individual's own opinion of what being fashionable means.Meanwhile, fashion consciousness regards the level of involvement and interest that an individual has in fashion styles and their own appearance (Nam et al., 2007).In this sense, the fashion consciousness construct is broader than 'fashion'.Individuals who are fashion conscious go one step further, seeking information on the latest trends and are active participants in co-creating value in the fashion production process (see Choi et al., 2016).Notably, fashion conscious consumers are influenced by the perceived value and fashion innovativeness of the garment, as well as trust in the fashion producer/retailer which, for the online retail context, regards website functionality, security, and the time saved through making online purchases (Escobar-Rodríguez and Bonsón-Fernández, 2017).Hence, fashion consciousness additionally involves an interpersonal or collaborative element between consumers and producers/retailers, as well as embraces social interactions between (potential) consumers of fashion whereby individuals seek the opinion of others.Iyer and Eastman (2010) further that fashion conscious consumersin contrast to consumers in generalspend more time shopping to refresh their collections.Thus, they are characterised by various psychological and social factors that impact not only their fashion consciousness in terms of associated feelings of self, but also their purchase intent in terms of frequency (see Kautish and Sharma, 2018;Vuong and Nguyen, 2018).
Millennials are inherently a fashion-conscious generation.Millennials view fashion as an artistic license or a status symbol by relating themselves to the fashion garments and brands they wear (Djafarova and Bowes, 2020;Samala and Singh, 2019).While millennials are presented as more willing to try novel materials and products than previous generations in the quest to look fashionable (Colucci and Scarpi, 2013), various studies suggest that millennials are particularly sensitive when choosing a fashion product or brand.For example, Rahman et al. (2020) find that brand awareness and brand consciousness are instrumental to the millennials' purchasing decision and that connecting to a particular brand and/or garment reinforces the millennials understanding of self (see also De Kerviler and Rodriguez, 2019;Leung et al., 2015;McNeill and Moore, 2015).Moreover, Casidy (2012) proposes that the perceptions of the brand and its price contribute to associated feelings of status within a particular social group for millennial consumers (see also Eastman et al., 2013).Overall, it appears that fashion consciousness in terms of hedonistic and/or social factors are extremely important for informing fashion purchase intent for millennial consumers (see McNeill and Moore, 2015;Rathnayake, 2011;Vuong and Nguyen, 2018) and the first hypothesis is: H1.Fashion consciousness positively affects millennials' fashion purchase intent.

CSR
Although considered a multidimensional concept which is difficult to define (see Carroll, 1991;Dahlsrud, 2008), CSR is often presented in the business literature as relating to the social, environmental and financial dimensions of Elkington's (1997) triple bottom line.Particularly, it regards a proactiveness by companies in terms of integrating societal, economic, ethical and sustainability concerns (see Sarkar and Searcy, 2016).For the fashion industry, CSR thus regards the ability to make profit whilst adhering to the norms or expectations of society in terms of social, environmental and ethical production.McNeill and Moore (2015) propose three groups of fashion consumers.While the first group 'self' is concerned with hedonistic needs and status, members of the second group 'social' are concerned with Table 1 Distinction between theories of planned and (un)planned consumer behaviour for this study.

Context
Psychological behaviour in various fields and disciplines building upon the theory of reasoned action (Ajzen and Fishbein, 1980) Online consumer behaviour

Motivations
Within online marketing, it is used to predict consumer behaviour (e. g.Londono et al., 2017;Sanne and Wiese, 2018) It is problematic to assume consumer behaviour is consciously tied to a planned process (or intentions) given that millennials are often hedonistic; subject to social pressure and may act spontaneously (see Alexander and Sysko, 2013).Therefore, the theory of (un)planned behaviour is motivated by the need to know more about why there is an intention-behaviour [see also Johnstone and Lindh, 2021]).For example, often incidental processes lead to sustainable purchases (see also Burke et al., 2014;Davies and Gutsche, 2016;Newton et al., 2015).Moreover, millennial consumers may purchase sustainably, but not be aware that the product they are buying is sustainable.

Behavioural process
Linear/deterministic Haphazard/unconscious L. Johnstone and C. Lindh their image in terms of peer evaluation.Meanwhile, the last group 'sacrifice' regards consumers who forego their sense of entitlement and self to reduce their social and environmental impact on the world.Arguably, this latter group prioritise the CSR work of fashion producers and retailers when making purchase decisions.While such categorises are useful for fashion retailers when marketing their products, there is the general assumption that consumer interest in sustainable fashion and ethical behaviour is growing (e.g.Joergens, 2006;McNeill and Moore, 2015), and therefore, the boundaries between groups are arguably becoming blurred.This is because consumer behaviour may be tied to status (i.e.self) and peer opinion (i.e.social), as well as reflective of individual sustainability values through an informed product choice (i.e.sacrifice).In this sense, fashion consumers may aim to look and feel good both on the outside and inside based on both the physical fashion garments (i.e. the tangibles) and internal sustainability values (i.e. the intangibles).
For millennials, Johnstone and Lindh (2018) suggest through their theory of unplanned behaviour that sustainable purchases in general often occur due to haphazard or unintentional processes, rather than intentional planned behaviour which incorporates all their values.Naderi and Van Steenberg (2018) further that millennials are primarily hedonistic, where social and environmental values are sidelined in decision-making processes.This stands against Fu and Liang (2018) who suggest that for eco-fashion, the personality traits of millennials in terms of their ecological and social consciousness are instrumental (see also Heo and Muralidharanm, 2019).Resultantly, extant literature paints an unclear picture of the degree of millennials' sustainability values in conditioning purchase intent as well as the interaction between hedonism and altruism/environmentalism; or particularly, between self and sacrifice (McNeill and Moore, 2015).
Beyond millennials, there is a stream of research which finds that the CSR practices of companies positively affect consumer behaviour and brand loyalty by reflecting the perceived personal values of consumers (e.g. Lee and Shin, 2010;Lee et al., 2012).Concurrently, the fashion industry is increasingly subject to CSR in terms of environmental (e.g.efficient and effective production processes that reduce waste and optimise resources) and social (e.g.fairness, equality and justice) sustainability practices (see Niinimäki et al., 2020).Nevertheless, Bandyopadhyay and Ray (2020) comment on the need to reconcile ethics and aesthetics in the fashion industry, claiming that fashion producers need to position strategies that embody both sustainability and fashion appeal.In this sense, the sustainability orientation of retailers is also important in informing purchase intent.
Together, the extant literature paints an unclear picture regarding how important the sustainability orientation of retailers is for informing millennials' purchase intent as reflective of their increasing sustainability values.Thus, the second hypothesis tests the assumption that sustainability is valued to some degree by millennials (see Heo and Muralidharan, 2019;Wang et al., 2018).More specifically, it asserts that millennials value socially responsive fashion retailers when making purchase decisions: H2.The CSR of the fashion retailer positively affects millennials' fashion purchase intent.

Importance of online influencers
Increasingly important for the millennial generation is being part of various online communities through social networks.Within such networks, social media influencers are considered essential for informing millennials purchase intent for fashion and beyond (Johnstone and Lindh, 2021;Chetioui et al., 2020).As Sashittal et al. (2015, p. 326) comment, "millennials speak a new language and value a new currency: social media speak and social media savvy".
Influencers are regarded as information disseminators that consumers follow in their social networking sites (see also Bakshy et al., 2011;Li et al., 2011;Khamis et al., 2017).Lokithasan et al. (2019) state that influencers have "established a reputation for themselves on social media" and whose success comes down to first, their entertainment value and second, their informativeness.Further still, influencers are distinct from 'traditional' celebrities in that their celebrity appeal is built through their following on social media, rather than another mainstream role.To this end, millennials view influencers as people like themselves (see Allsop et al., 2007;Mangold and Smith, 2012) who allow the millennials to position themselves socially in terms of consumption patterns through an online 'celebrity subculture' (Hamilton, 2010).Here, Lee et al., (2021) find that influencer following is built through motivations tied to his/her authenticity, consumerism, creative inspiration and envy.
For companies, social media posts (i.e. on social media sites, vlogs, blogs etc.) shared by influencers help increase brand awareness and preference, as well as subject matter expertise (Chopra et al., 2020).Chopra, Avhad and Jaju (2020) argue that influencer opinion is even deemed more valuable than the millennial's immediate peersor traditional word of mouthin building, for example, brand-loyalty (see also Purani et al., 2019).Schouten et al. (2020) further that consumers trust influencers more than traditional celebrity endorsers.Here, consumer trust is built indirectly through social media exchanges that move beyond the traditional buyer-seller dyad (Johnstone and Lindh, 2021) to reach a wider consumer base than conventional offline channels (see Johnstone and Lindh, 2018).Nevertheless, Martínez-López et al. (2020) highlight that consumer trust in the influencer is reduced if the post is perceived to have a commercial orientation.This suggests that posts must appear to genuinely reflect the influencer's opinion to be successful (i.e.be perceived as authentic).
The fashion industry deploys influencers as communication tools (Esteban-Santos et al., 2018) as social media advertising becomes increasingly important (Geissinger and Laurell, 2016;Pedroni, 2016;Wiedmann et al., 2010).As a high involvement product category which attracts consumers through personal identity and status among others, fashion engages consumers to converse online (Gu et al., 2012;Iyer and Eastman, 2010).Particularly, Wolny and Mueller (2013) find that commitment to brand and fashion involvement inspires consumers to talk about fashion and interact with fashion brands.They also find that consumers who have a higher need for social interaction (e.g.millennials) more frequently engage in such online conversations and various studies highlight the importance of influencers in fashion purchasing for both male and female consumers alike (e.g.D 'Souza et al., 2015;Wiedmann et al., 2010).Therefore, the following hypothesis is: H3.The influence of influencers positively affects millennials' fashion purchase intent.

Interaction effects
The discussion thus far suggests that fashion consciousness, social media influencers and CSR all directly influence the purchase intent of fashion garments for millennial consumers.While the roles of fashion consciousness and influencers on purchase intent for millennials remain relatively uncontested in the extant literature, the role of CSR on purchase intent for millennial consumers is less clear.Building upon the theory of unplanned behaviour and trust through intermediaries in the online retail context, it also becomes of interest to explore if and how millennials purchase behaviour is furthermore affected through intermediary pathways.Thus, two further assumptions are built into the research model regarding the role of online influencers for millennial consumers in the relationship between CSR/fashion consciousness and purchase intent.These assumptions are motivated in terms of the interaction effects in the increasingly heterogeneous online retail context where linear pathways to purchase intent are becoming furthermore obscured (see Johnstone andLindh, 2018, 2021).
First, the latest fashion trends tend to be followed by millennials who constitute one of the most 'self-conscious' consumer groups in terms of how they are perceived by others, namely peers.This is where McNeill and Moore's (2015) notion of 'social consumers' -who are concerned with the opinion of peerscomes into play.Leung, Yee and Lo (2015) find that the fashion consciousness of such generations is positively affected through the media.Consequently, the use of influencers as 'celebrity endorsers' helps create brand image and loyalty (see McCormick, 2016).Previous studies suggest that millennials follow online fashion blogsas one influencer outletfor information on products or brands, in addition to entertainment purposes (Esteban-Santos et al., 2018).The value attributed through this form of 'online review' by the influencer is subsequently based on how the millennial views the credibility and trustworthiness of the influencer as intermediary (Hahn and Lee, 2014).This is determined through social interaction (Esteban- Santos et al., 2018).Thus, if the millennial views the online influencer as credible and trustworthy, the millennial consumer is more inclined to view, search for or even buy the fashion product/brand reviewed by the influencer.As such, the influencer serves an intermediary function in the relationship between the fashion retailer or brand and the consumer: H4.The effect of fashion consciousness on millennials' fashion purchase intent is strengthened through influencers as intermediary marketing channels in the online retail context.
Second, given that it remains unclear the extent to which CSR as reflective of individual sustainability values directly promotes purchase intent for millennial consumers, the role of influencers as intermediaries to promote sustainable purchases is also of interest.Previous studies highlight that online interpersonal relationships positively affect green purchase intentions (Bedard et al., 2018); whether 'intentional' or not (Johnstone and Lindh, 2018).Johnstone and Lindh (2018, p. 130) argue that online influencers may be effective marketing tools for the "internalisation of sustainable [buying] behaviours", more so than those based solely on moral arguments or axiological concerns by the millennial consumer.Thus, influencers can be considered important for fashion brands or retailers to promote not only increased loyalty and sales, but also their internal CSR work and thus, sustainable futures for all.This is because millennial consumers are increasingly influenced by online influencers, who are deemed more important than conventional WOM channels (Chopra et al., 2020).Here, trust is tied to how the millennial views the opinion of intermediary actors, rather than (solely) on the sustainability of the retailer, brand or product.Consequently, the reversal of socially and ecologically detrimental consumption patterns can be mitigatedeither consciously or unconsciouslythrough influencers as intermediaries promoting a sustainable self-identity of the millennial:

H5. The effect of CSR on millennials' fashion purchase intent is strengthened through influencers as intermediary marketing channels in the online retail context.
Together, these secondary hypotheses suggest that the effects of CSR and fashion consciousness are stronger if online influencers are deployed by fashion retailers as strategic marketing tools.This, in turn, may translate into increased fashion purchases in both the online and offline retail context for millennial consumers.

The research model
Through its assumptions, the research model (Fig. 1) anchors its hypotheses in the theory of unplanned behaviour by testing both the direct and indirect pathways (specifically influencers as one type of intermediary in the online retail context) promoting millennial purchase intent in the online retail context (see Johnstone and Lindh, 2018).Here, purchase intent is used to provide a rough estimation of consumer buying behaviour, as it relates to the "individual's conscious plan to make an effort to purchase a brand" (Spears and Singh, 2004 p. 56) after making evaluations in terms of satisfying their needs and wants (Vuong and Nguyen, 2018).Notwithstanding, in extension to previous research and the previous definition, the model proposes that the very action of purchase intent may well be unconscious (or unintentional); in this case, through millennials interacting with influencers.Thus, the research model builds upon the proposition that influencers are increasingly important communication channels or conduits for the online fashion retail context (see Djafarova and Bowes, 2020;Wiedmann et al., 2010), and could be used by retailers to minimise the sustainable-fashion gap (Johnstone and Lindh, 2018).By loosely coupling these assertions to the role of trust, it presents the relationship between consumers (namely millennials) and brands as an interactive one; reinforced through intermediaries such as online influencers as e-communications channels (see Johnstone and Lindh, 2021) which may be used in various ways to support the brand/retailer in the transition to more sustainable business models.

Method
This study aims to better understand how sustainable business models can be achieved and what this means for the assumptions of extant marketing theories in an increasingly online retail context.Notably, it frames this considering the sustainable-fashion paradox for millennial consumers, given the tensions within said generation regarding axiological and hedonistic concerns tied to fashion consumption.To achieve such aims, a standardised survey which involved a two-stage analysis was designed.First, regression and reliability tests were conducted with alpha on the measurement items.Second, a more explorative approach was performed in LISREL to test the measurement items from the developed research model in this study, particularly in terms of the interacting effects suggested.The reason for these two steps was that although previously tested items were used for many of the suggested constructs in a regression analysis, the subject area regarding influencers (see Martínez-López et al., 2020) and the sustainability-fashion paradox (see e.g.Bandyopadhyay and Ray, 2020;Rolling and Sadachar, 2018) is rather new.Thus, the constructs require refinement.Particularly, the idea to assemble the constructs in the research model in this study calls for scrutiny of their fit together.The result of this second stage analysis was a trimmed, reliable and useful measurement scale/set of items for fashion consciousness, CSR, influencers and fashion purchase intent which can be used in future research.

The survey and data
To achieve the data needed for analysis of a disparate phenomenon and fairly complex model, a survey was conducted.This sets high demands on the items and their scales (Boateng et al., 2018).Further, in terms of the two-part research design, the data needed to permit regressions and test suggested paths, as well as fit into the structural equation modelling (SEM) (Martínez-López, Gázquez-Abad and Sousa, 2013).Therefore, for this study, the recommendations for SEM by Jöreskog and Sörbom (1993) are employed.The survey was designed and administered to an international dataset of consumers during 2019 and 2020. 1 The survey was available in English, Spanish and German, and back translations were conducted to ensure the meaning of measurement items were consistent for each of the languages.Given that it is impossible to define the population of online fashion consumers internationally as the number of people who purchase fashion online is unknown, this survey was distributed through a combination of convenience, snowballing and purposive sampling techniques.Notably, respondents canvassed for completing the initial survey were not targeted in terms of particular demographics and control variables were later applied.
Respondents were initially canvassed through personal channels (i.e.social media platforms, SMS and emails).Further, a shareable weblink was embedded into the survey's design for respondents to forward it on to others in their social networks.Finally, the survey was also shared on social media groups dealing with fashion when permitted.Together, these approaches resulted in 733 respondents from 56 countries over a six-month period (see Appendix A).Of them, 448 observations were categorised as 'European millennials' which are controlled for in the regression analysis.

Variable measures and items
Questions were constructed to test the research model on purchase intent as the dependent variable, and fashion consciousness, CSR and influence from influencers as the independent variables.To permit multiple linear regressions, a seven-degree scale was used (1 -disagree, 7completely agree).While most of the variables regard single groups of measurement items, fashion consciousness and CSR were further subdivided to reflect trends in the extant literature as described below.
Purchase intent regards the likelihood of an actual purchase occurring.It is based on action, rather than attitude (see Spears and Singh, 2004) and is assumed as the outcome in the research model guiding this study.While attitude informs the intent to purchase, the measurement items developed for purchase intent as the dependent variable regard the act of purchasing itself.At the same time, purchase intent can only offer an estimation of likely purchases.This is because the actual purchase behaviour in terms of purchases made on fashion garments cannot be determined by the survey results itself.Therefore, purchase intent is used as a proxy variable which provides estimations of future purchasing behaviour.
Inspired by Yang et al. (2014), the following items were used to measure purchase intent on a seven-point scale from definitely/definitely not: 'I intend to buy more fashion garments soon'; 'I believe I will buy more fashion products in the future than I do now'.These items have the dimension of time (i.e.fast fashion and its seasonal basis) as well as plurality (i.e. in terms of multiple fashion garments) in order to capture the notion of fast fashion purchasesin this casefor the millennial sub-group.Nevertheless, they refer to purchase intent in general, rather than specific to the online retail context.To this end, additional items were added to reflect purchase intent for the online context over time (see Table 2).These were constructed for this study based on summarising ideas from the extant literature, as well as influenced by Yang et al. (2014).Thus, the measurement items for purchase intent reflect both online and offline purchases given that although consumers may be influenced to purchase fashion garments online, the actual purchase may occur in a 'traditional' offline store.
While there is an apparent connection between fashion consciousness and purchase intent proposed in extant research on millennials (e.g.McNeill and Moore, 2015;Vuong and Nguyen, 2018), the dimensions that are included within the fashion consciousness construct are more diverse.Gould and Stern (1989) developed the 'Fashion Consciousness Scale' to measure the individual's level of fashion consciousness.It is based on self-consciousness in general and the daily concept of fashion which individuals have.Although having 32 measurement items, the scale has been used and simplified since then (e.g.Casidy, 2012;Kautish and Sharma, 2018).Furthermore, Rieke et al. (2016) suggest that perceived, social and environmental factors influence the intent to purchase fashion garments for female consumers.While the perceived factors regard confidence, body image perceptions and personal preferences, the social factors include celebrity endorsement, consumer morals and beliefs, and the environmental factors regard friend and family influence and appropriate choices for the elements.In this sense, the environmental factors relate to culturally and environmentally appropriate choices within a particular context.The development of the previous scales and measures suggest that there are various internal and external factors which affect fashion purchase intent.Therefore, the 16 items used to capture fashion consciousness in this study -in light of increasing fast fashion trends -regard seasonality, trends, personal identity, peer opinion, celebrity status and branding among others.This builds upon the previously mentioned works but groups them into social, personal and environmental categories for the purpose of this research.
CSR is captured in six measurement items which group CSR into two category groups, namely: personal expectations/values by the consumer and CSR communication of the company.These groupings consequently link the individual and organisational values respectively.Particularly, they reflect the previous discussion on sustainability values as internalised by consumers on the one hand, and expected of by the retailer on the other.
For CSR communication, this regards items that capture CSR in terms of its third-party verification, transparency and consistency (see Kim and Ferguson, 2018).For personal expectations/values, this regards items that capture the respondents' values in terms of social contribution (i.e. consumer protection and welfare), environmental protection and local community contribution (cultural activities and local community development) (see Lee and Shin, 2010).
Finally, influencer influence relates to an eWOM channel whereby the millennials intent to purchase is affected by the influencer as online 'reviewer' and 'promoter' (see Johnstone and Lindh, 2021).The influencers' influence is contingent on his/her followers' trust and value of the influencer's opinion of the fashion garment.Consequently, the seven measurement items developed for influencer influence regard the relationship between the influencer and the millennial, and social network capital, namely: interpersonal trust, information exchange, online community involvement and shared values, among others (see Krishen et al., 2018).
As an overview, the items that were used in this research are displayed in Table 2.The Cronbach's alpha values in Table 2 confirm the strong reliability of the constructs' internal consistency, except for the dimension of fashion consciousness connected to environment (FCE).This is somewhat lower (0.493) but still acceptable (Nunnally, 1978).Nevertheless, to compound the different variables and create new variables after the testing with alpha, a factor analysis with principal component (varimax rotation) was used.This resulted in new variable measurement items for purchase intent (PI), influencers (INFL) and fashion consciousness (FC) which are later used in LISREL analysis.2Thus, through its two stage analysis, the study helps refine constructs for future research in the area.

Controls and assumptions
This study focuses on the millennial sub-group of consumers and most of its respondents were from the European regional bloc.Therefore, the chosen control variables regard age (i.e.millennials as those born between 1981 and 1994 [Mangold and Smith, 2012]) and geography (i.e.European citizens), which were applied in the regression analysis to test the research model.Notably, gender was not included as a control variable for the following reasons.Although females were the most represented group in this study, 3 Jin and Ryu (2020) suggest that studies that differ between genders are imperfect as they assume no variation between genders.Rather, they argue that in terms of influencer marketing, there is the need to 'tone down' the role of gender in study designs and not view each group as an 'either-or'.Moreover, additional testing found no conclusive differences between gender on the research model.
Beyond the motivation that millennials consumers are perhaps the most important for the fashion industry (see Gerardo, 2018) and are most influenced by influencers to inform fashion purchase intent in comparison to other consumer groups (see Pedroni, 2016;Wiedmann et al., 2010), geography was also presented as a control in this study.While indeed there may be differences between markets in terms of physical or psychic distance in the online retail context (see Safari et al., 2013), the question of whether a market is global or multi-local remains.a The ability to exchange information on fashion garments with a social media influencer is important to me.I am more likely to buy a product if an online influencer reviews it positively.I am more likely to like a brand if an online influencer reviews it positively.
a Items that are omitted from the constructs in the second part of the analysis, as they do not fit in the measurement model. 3The fact that women were more represented in the dataset is thought to be down to the receptiveness of women in completing surveys/questionnaires for research (see also Sethi et al., 2018).Parker, Hermans and Schaefer (2004) found as early as no significant differences between Japanese and American consumers when it comes to fashion consciousness, but differences between 'less' developed and developed countries; thus, implying that fashion holds some international character which is based on the socio-economic status of the country.More recent studies imply that the Internet may also hold an 'international' character in terms of the consumers it reaches and that there are benefits from international datasets (e.g.Anastasiadou et al., 2019).
Based on these assumptions, the sample was first divided into millennials and other age groups.This was to explore the significance of age differences in terms of the research model which poses that following influencers is age-related.Second, somewhere between focusing on single country analyses for fashion purchase intent (e.g.Casidy, 2012;Kautish and Sharma, 2018) and treating consumers as homogenous internationally, the control variable of Europe is applied in this study to explore bloc-wide patterns. 4The decision to focus on European millennials was because most respondents were located within the European context.It furthermore builds upon the assumption that online influencers can reach beyond their immediate contextual environment and that European consumers inform an international dataset with relatively similar fashion styles due to the rise of high street fashion brands that are available in most European towns and cities, among others.
Applying these two control variables in the regression analysis helped test the assumption that online fashion marketing and the fashion industry in general arguably have an international or regional character for the millennial sub-set of consumers.This is embedded into the assumption that the worldwide webin generalas well as fashion trends and sustainability issues know no geographical boundaries, and products are increasingly delivered internationally.Moreover, it also permits the effect of influencer influence on purchase intent to be tested as a marketing strategy connected to age.
Thereafter, testing in LISREL included only those respondents categorised as European millennials as well as excluded the measurement items from the regression analysis that did not fit the measurement model (those highlighted in Table 2).This means that the data in the structural model are 'valid' in terms of discriminant and convergent validity given that the significance levels are the same and the strength of paths are shown as t-value and factor loadings (standardised solution), and the load for each item to its construct is assessed by R2 (see Table 3, where it shows they exceed 0.2) as suggested by Martínez-López et al. (2013).To be sure there are no threats to the validity and reliability of the model, an examination of t-values exceeded 1.96 and the factor loading (completely standardised solution) exceeded 0.3 (Holm et al., 1996).The nomological validity (i.e. the internal logic of the work) presents the two parts of the analysis as consistent with each other.To further test this for the structural model, several fit indexes were examined, shown in Table 3 (namely, AGFI and GFI [Hayduk, 1988], CFI and RMSEA [Byrne, 2001] and TLI (NNFI) [Bentler and Bonnet, 1980].Particularly, both the regression analysis and structural model show the importance of influencers for millennials' purchase intent in the context of fashion purchasing.

First stage analysis -regressions
To investigate the potential paths between the investigated concepts and evaluate the research model guiding this study, a multiple regression was first performed in SPSS to find significant paths between the constructs.Table 3 displays the β coefficients t-values and significance to indicate the strength and security of the paths.The significance levels are indicated by stars and show the "certainty" on the strength (β and tvalue) of the path.A dagger ( †) means that the level of significance is not sufficient to be conclusive, but that the path cannot be dismissed either and is thus, is a call for further testing/research; herein the secondary LISREL analysis which follows.Within Table 3, Models 1-4 concern the independent variables separated in terms of the thematic categories of the measurement items.In this sense, there are 'five' independent variables presented: three on fashion consciousness (personal [FCP], social [FCS] and environmental factors [FCE]), two on CSR (CSR communication [CSR1] and personal values [CSR2]), and the variable of influencer influence [INFL] which is independent in Models 1-4 with purchase intent [PI] as the dependent variable.Meanwhile, Models 5-7 present INFL as the dependent variable.For each test, age (millennials and non-millennials) and country (European and non-European) are used as controls.

Purchase intent as the dependent variable
Model 1 indicates that both FC and INFL affect PI, but that CSR has a weak (although negative) significant effect.Age or country have no effect in this model.Meanwhile, Model 2 tests FC and CSR on PI (without INFL).When excluding INFL, the effect of FC is strengthened and the negative effect by CSR is reduced, i.e., it is only significant on the "dagger level" which calls for further research.This means that when influencers are not considered, CSR has no significant effect on PI, but FC has a strong effect.It furthermore appears through Model 3 that the two sub-constructs of CSR1 (communication) and CSR2 (personal values) both have significant effects on PI, although CSR1 is negative and CSR2 is positive, but weak.This means that the individual values associated with CSR do affect PI, although to a small degree.However, the CSR communication work of the fashion retailer in this case has a negative relation to PI, suggesting that this is not something that would make the respondents buy more.Additionally, Model 3 indicates that the relationship between CSR and PI could be affected by both age and country, thus meriting more testing.Finally, Model 4 shows the FC subcategories and that interestingly, FCE has no significant path to PI, while both the social (FCS) and personal (FCP) do.

Influencer influence as the dependent variable
Model 5 suggests that CSR1 on INFL is not significant.However, the dagger implies that the role of CSR communication by the fashion retailer in relation to INFL merits further study.Meanwhile, CSR2 has no significant effect on the influence of influencers in this model.Model 6 implies that the overarching FC and CSR variables both have a significant impact on INFL.However, when FC is subdivided into its three subconstructs, FCE is not significant (Model 7).This means that CSR must be built up by both communication and personal values to have a positive relation to INFL on one hand, but that the environmental issues associated with fashion consciousness (FCE) have no bearing on the role of influencers.Note that for Models 5-7, age but not regional bloc makes a difference when it comes to influencers.This supports the idea that the millennials behave differently in fashion purchasing in comparison to non-millennials, but also that millennials exhibit some sort of international character in terms of FC and CSR.Together, these findings call for further analysis of the millennial sub-group of consumers.

Summary of regressions
The results suggest that fashion consciousness and influencer influence matter for fashion purchase intent.Interestingly, CSR's role on purchase intent is disparate, as it is composed of negative and positive impacts.Regarding influencers, it appears that both communication and personal sub-categories of CSR must be considered to have an effect.This finding calls for further investigation.CSR as antecedent of purchase intent is negative when measured as communication, but slightly positive measured as personal values.This contrasts to its relation to influencers when they are dependent.Finally, age seems relevant for the propensity to follow influencers but not necessarily geography.Notwithstanding, the environmental aspect of fashion consciousness does not impact purchase intent or influencer influence, even though it may form part of a larger construct that includes social and personal aspects.

Second stage analysisindirect effects of influencers
In alignment with the research model, the regression analysis suggests not only direct, but also indirect effects on PI through INFL.Nevertheless, the regressions also suggest that further tests are required to test initial assumptions through refined constructs.Thus, the assumption that INFL play a role in the relationship between FC/CSR and purchase intent for European millennials (n = 448) are tested in LISREL on the slightly modified constructs which fit as indicators5 (see Appendix B).Table 4 presents the results of this testing where the models and coefficients of the paths offer reliable indications which further validate the constructs.
Five models have been explored and fit to the data.Model 1 confirms that FCP directly impacts PI and is also mediated by INFL.Meanwhile, Model 2 shows that FCS has no direct path to PI but there is an indirect effect by influencers.Model 3, however, only has one significant path from INFL to PI, which rejects the idea of FCE having any effect on PI, even through influencers.
Meanwhile, CSR in the context of influencers and fashion purchasing yields some interesting results.Model 4 suggests that CSR1 (communication) has no direct impact on PI.Nevertheless, the effect of CSR1 on INFL is positive and significant, as is the path from INFL to PI.This means that even if CSR1 has no positive or significant effect on PI, there is a positive, indirect, effect when influencers are included.Meanwhile, CSR2 (personal values) has a direct negative impact on purchase intent.However, it is positive for influencers, and shows as in the case of CSR1 a mediating effect by the influencers.Finally, Model 5 shows that even negative fashion-related perceptions can be turned into something positive and lead to purchase if there is an influencer involved given the effect of influencers is significant.

Revised research model
The results of the structural model indicate some interesting pathways for millennial consumer behaviour in the online fashion retail context, which results in a revised research model (Fig. 2).Moreover, the results of the two-step analysis help refine the initial research model and its constructs which can be re-worked into future research.
It appears that influencers are especially important for CSR and the social aspects of fashion consciousness to positively affect millennials' purchase intent.That is, without influencers, CSR and the social aspects of fashion consciousness (i.e.friends, family and peer opinion) would have no direct effect on the European millennials' purchase intent.This latter finding is especially interesting as it suggests that millennial consumers base their fashion purchase intent more on celebrity influencers than peers with whom they may have physical social relations with.The model further emphasises that influencers condition the relationship between CSR, FCS and purchase intent.Influencers also strengthen the effect of the personal factors (i.e.primarily in terms of hedonistic values through the refined construct) related to fashion consciousness on millennials' purchase intent.In this sense, it appears that influencers impact the effect of the applied strategies on the purchase intent, shown in the analysis as the indirect effect is significant in all models but Model 3.This leads to the revision of the initial research model.

Concluding discussion
This study builds on testing the inherent paradox between increased sustainability values and fashion consumption for millennial consumers as the biggest and most fashion-conscious consumer group.Particularly, it aims to tease out if and how millennial consumers are affected to buy more sustainably through influencers as an online marketing strategy, as well as if and how the hedonistic needs of fashion-conscious millennials can be reconciled with their internal sustainability values through such a strategy.To this end, the following discussion bridges these corporate and consumer sides by relating the findings of this study to prior literature in the area.

Millennials -A more sustainable generation?
The detrimental social and environmental impact of the fast fashion phenomenon has led to a move towards slow fashion and a 'conscious retail model' in recent years (Jung and Jin, 2016;Pookulangara and Shephard, 2013).Nevertheless, the degree to which sustainability concerns have been integrated into millennials' purchase intent remains unclear (e.g.Bandyopadhyay and Ray, 2020;Rolling and Sadachar, 2018).
The findings of this study question the assertion that millennials constitute a more sustainable generation which is reflected in their purchasing behaviour (see Heo and Muralidharan, 2019;Wang et al., 2018) and the decision to buy from brands and retailers that align with their 'sustainability' values (see Henninger et al., 2017;Miotto and Youn, 2020).Rather, it appears that when it comes to fashion, other factors may be more important and that millennials are not (yet) truly emerging sustainability consumers (see also Joergens, 2006).
For the millennials in this study, personal factors associated with fashion consciousness were the most instrumental for informing purchase.These regard the fashion brand chosen and the millennials' perceived image, rather than any intrinsic concern for sustainability or expectations on the fashion retailer for having sound CSR practices.This first implies, in accordance with previous research (see Carrington et al., 2014;Johnstone and Lindh, 2018;Kumar et al., 2017), that there is a value translation gap between the internal sustainability values of consumers (in this case millennials) and the actual purchase intent (in this case of fashion garments).It also stands against the assumption that CSR communication or sustainable fashion marketing itself is integral in informing consumer opinion (Schmeltz, 2012;Lundblad and Davies, 2016).Rather, the purchase behaviour of the millennials in this study is primarily motivated by their hedonistic needs in terms of perceived image (see McNeill and Moore, 2015).
That being said, the regression analysis offers further subtle nuances which suggest that to some degree, the personal values associated with 'sustainability' are more important for millennials than how the fashion 55.02 (36)  59.88 (36)  51.69 (37)  54.52 (33)  51.60 (33)    L. Johnstone and C. Lindh retailer communicates its CSR work.This latter finding implies that a personal or moral obligation felt to act sustainably or 'sacrifice' (see McNeill and Moore, 2015) to reduce their impact on the world, may well be emerging.Although, this foregoing of -what McNeill and Moore (2015) call -'self' for a broader social and environmental good per se, is not quite there yet for the millennials in this study, even if others suggest that social and environmental commitment is instrumental for sustainable consumer behaviour in the fashion industry (e.g.Mishra et al., 2021).This is because looking good on the outside still takes precedence for most millennials in this study given that the internal values associated with 'CSR' are not strongly reflected in purchase intent as action.It furthermore suggests that it is only a sub-group of millennial consumers that can be regarded as having aligned intention and behaviours when it comes to sustainable purchase intent which supports McNeill and Moore's (2015) assumptions that consumers can be categorised into distinct groups.

The role of the fashion industry in supporting a sustainable fashion transition for millennials
The extant literature presents the rise of fast fashion and a growing fashion consciousness for millennials as the biggest consumer group in conflict with their internal sustainability values.However, the findings of this study reveal a more complex picture where the online marketing strategies adopted by fashion retailers are instrumental to the solution.
On the consumer side, there is the assumption associated with the slow fashion movement that consumers increasingly demand the fashion industry to work more with CSR to ensure sound working conditions and minimise environmental destruction (Henninger et al., 2016).This suggests that changes made in the fashion industry are made in response to customer concerns.However, the findings of this study suggest that millennials, as a sub-group of fashion consumers, are not necessarily driving the transition of the industry to a more sustainable one.Rather, the push is coming from the socio-political context at large, which includes governments, trade unions and other consumer groups or sub-sets of millennials, among others.This is because the main driver for most millennials' purchase intent in this study regards hedonistic factors rather than any altruism concerned with the social and environmental issues within the fashion industry.
Therefore, on the retailer side, the findings indicate that fashion retailers still have the ultimately responsibility when it comes to promoting sustainable purchases for most of the European millennial consumers studied.Here, the responsibility focus is on the retailers through the revised research model wherein the use of influencers downplays the role of millennial values in the transition to a more sustainable fashion industry.Ultimately, it becomes the role of the industry to promote a sustainable change in consumer behaviour (see also Lee et al., 2020).In this sense, the study inadvertently places responsibility on the fashion industry to move to a sustainable business model (see also Jung and Jin, 2016;Todeschini et al., 2017).But, not only that; fashion retailers and marketers thereafter have a duty to convey that model to consumers of fashion (in this case millennials) through various strategic marketing strategies (in this case online influencers), whether explicitly or not marketed as sustainable garments (see Johnstone and Lindh, 2018).This is because not all millennials' ideals or values on sustainability are reflected in practice yet there is still pressure on the fashion industry to be sustainable.

Reconciling the sustainability-fashion paradox for millennials through influencers
Whether or not the internal values connected to CSR are genuinely increasing within millennials remains unclear through this study, but what is clear is that both the personal values associated with CSR and the fashion retailer's CSR work are not enough to affect the millennials' purchase intent of fashion garments.While there is some indication through the regression analysis that the personal values associated with CSR might affect fashion purchase intent for a small sub-group of the millennials studied, it appears that reconciling the paradox between the increasing fast fashion phenomenon and sustainable futures for millennials more broadly comes down to the fashion retailers and marketers.Particularly, the results highlight the need to employ influencers in the promotion of fashion garments for millennials to make more sustainable fashion choices, whether conscious or not in terms of the sustainability appeal of such products (see also Johnstone and Lindh, 2018).This implicates a more complex relationship between the producers, retailers and consumers than previously indicated for the online retail context.Specifically, it emphasises the necessary role of other online actorsin this case influencersas marketing intermediaries that can help promote not only fashion-conscious decisions, but also sustainability values and beliefs for millennials.It finds that influencers are not only channels to promote fashion consciousness as Leung et al. (2015) imply, but also to (un)consciously promote sustainable product consumption (Johnstone and Lindh, 2018).
Importantly, embedded into the deployment of influencers as an online marketing strategy to aid in the transition of a sustainable business model from the consumer side is the concept of trust.While it appears that the internal values associated with sustainability/CSR for millennials do not translate into action in this study (see Carrington et al., 2014), the online interpersonal relationships between the millennial and social media influencer appear to drive sustainable purchase behaviour.This suggests an openness to and direct trust in influencers for millennials in the online retail context, furthered by the finding that influencers are more important than friends, family and peers in informing fashion purchase intent.It thus builds on notions of alternative pathways conditioning consumer behaviour online.Here, trust is not directly between the retailer and consumer as often implied (e.g.Mukherjee and Nath, 2007;Pappas, 2016), but rather is built within social exchanges between intermediary actors which can be multiple in forms (i.e.review website, influencer, online peers etc.) and the consumer (Johnstone and Lindh, 2021).This means, in support of Johnstone and Lindh (2018), that the linearity commonly asserted in theories of planned behaviour should be reconsidered for the increasing online retail context.Indeed, while the use of influencers to boost and inform fashion purchase intent as well as the fashion retailers' CSR work may seem to counterargue the slow fashion model on the one hand, influencers can help in the move towards more sustainable product consumption choices, whether conscious or not for their followers.Rather than 'slowing' the fast fashion model down dramatically, the findings of this research present a sustainable approach in terms of economic, social and environmental dimensions to increase sustainable production practices.And, in this sense, as Henninger et al. (2016) suggest, slow and fast fashion are not in competition, but the question rather becomes how to 'sustainably sustain' fashion consumption.

Theoretical implications
This work contributes to prior research and practitioners by providing a more nuanced understanding of the sustainable-purchase gap for the fashion industry in an increasingly online retail context.Theoretically, it builds upon the limitations in assuming consumer behaviour as a linear process by suggesting that online retail is characterised by multiple pathways that thus, requires the expansion of extant behavioural theories which fail to comprehensively address such relational complexity.More specifically, moving beyond a production and/or consumer focus, this study presents fashion retailers as currently at the heart of the transition to more sustainable business models in terms of promoting sustainable consumer behaviour (an inside-out approach).As a whole, the study provides contributions to consumer marketing, and especially the emerging areas of influencer marketing L. Johnstone and C. Lindh (see Martínez-López et al., 2020), international online markets (Anastasiadou et al., 2019) and the sustainable-fashion paradox for retailers (Bandyopadhyay and Ray, 2020;Islam et al., 2020;Mukendi et al., 2020).It also contributes to the emerging sustainable-fashion literature in terms of marrying the aforementioned marketing literatures for the fashion industry as the empirical context (Bandyopadhyay and Ray, 2020;Bertram and Chi, 2018).
The findings also reveal that trust is embedded not necessarily built into the direct relationship between the retailer's online presence and consumer as previously suggested (e.g.Irshad et al., 2020), but rather through the online social relationships between the (potential) millennial consumer and the influencer as intermediary.It proposes that influencers are paramount information disseminators in the online retail context for European millennials for something so 'highly personal' as fashion.Thus, the study contributes to an emerging stream of research that aims to better understand consumer trust through social media marketing as the mechanism that positively affects purchase intent.These contributions can be built on in future research given that the COVID-19 pandemic has arguably pushed more consumers online.
More specifically, as an outcome, the study presents a refined research model which emphasises the complexities in online purchasing behaviour for millennials in terms non-linearity and unintentionality.Thus, it offers empirical evidence to support Johnstone and Lindh's (2018) theory of (un)planned behaviour.Nevertheless, there is the need to test and build on the revised model and constructs proposed in this study.Particularly, even though the retailers are presented as driving sustainable behaviour in the current model rather than the consumers per se, that is not to say that millennials may not be the ones driving the sustainable transition of the fashion industry more in the future (i.e. an outside-in approach).Therefore, the models, theories and pathways presented in this study may require adaptation as the online retail context develops.
Future studies can further refine the fashion consciousness variable for today's market beyond the application of Gould and Stern's (1989) original scale.Moreover, it may be of interest to draw out personality differences in terms of the research model results and, specifically, the fast fashion paradox.For example, it could be of interest to expound upon how millennials' fashion values are affected by various psychographic variables which come down to personality type, rather than gender (see also Jin and Ryu, 2020).This would involve a study design with further questions to assess the type of millennial answering the questions.Here, a more nuanced understanding of the sustainability intention-behaviour gap would be better understood (see Carrington et al., 2014).Further, the controls embedded into this model suggest few differences between region (i.e.European geography and fashion) and gender which future studies can investigate further.It would also be of interest to add in other intermediary (f)actors for the online retail context in addition to influencers (e.g.comparison websites, customer reviews, online peers).In this sense, future research regards building increasingly complex models to test the interactions on the online retail context for promoting sustainable purchase intent as an emerging area of interest.
Additionally, given that purchase intent only offers a proxy-variable of the actual fashion purchasing behaviour, future research would benefit from a qualitative approach.This would provide a more detailed understanding on the effect of CSR and influencers on sustainable purchasing behaviour from the millennials point of view perhaps through the form of focus groups to supplement the findings of this initial research.
Finally, although this study concentrates on the fashion industry, its findings in terms of models of unplanned behaviour and consumer trust may also be transferrable to other product segments and/or age cohorts.Therefore, future studies could build on the model for other product categories and/or generations.

Practical implications
The findings of this study offer some strategic guidance to fashion retailers on how to increase purchase intent whilst supporting the transition to a more sustainable business model.This is important given the growing socio-political pressures on the fashion industry to move beyond fast fashion business models and reflect increasing axiological concerns for the social and natural environment.
Arguably, this study contributes to slowing fashion down and a conscious retail model from the retail side (e.g.Jung and Jin, 2016;Leung et al., 2015;Pookulangara and Shephard, 2013).Specifically, it implies that fashion retailers not involved in social media marketing may be at a disadvantage in not only attracting millennial consumers, but also (un)consciously promoting sustainable purchase behaviours within this group.Given that online platforms are underutilised for CSR work (Kent and Taylor, 2016), it proposes that fashion retailers employ influencers to help market sustainable fashion choiceswhether conscious or not for the millennial consumer.The use of influencers by fashion retailers is important given that this study finds that agebut not country (see also Anastasiadou et al., 2019) is a decisive factor in the deployment of influencers as a marketing strategy.
Importantly, while fashion consciousness is still perhaps the most valuable antecedent of purchase intent for the millennial consumers in this study, there is potential in fashion producers and retailers moving towards a balanced approach to production and a sustainable business model (see Ozdamar Ertekin and Atik, 2015 ).This suggests that not only looking good on the outside but feeling good on the inside about sustainable purchases will become increasingly important in the following years and fashion retailers must work with bridging the value-action gap between sustainability values and hedonism when it comes to fashion consumption.

Declaration of competing interest
None.

Table 2
Measurement items, construct and alpha per construct.The comfort and fit of fashion garments are important to me. a The quality of fashion garments is important to me.
a a Strangers' opinions of my fashion choices are important to me.How people online view my fashion choices is important to me. a The latest fashion trend is important to me.

Table 4
Paths of the five tests for mediation.

Appendix B. Refined constructs for future research
buy more fashion products soon.I intend to keep buying fashion products on the Internet I intend to keep buying fashion garments from the online retailers I buy from today.I intend to buy fashion garments from new online retailers in the future.Fashion brands are important to me.The fashion garments I buy are important for my image.Fashion garments that look good are important to me.Social(FCS) My family's opinion of my fashion choices is important to me.My friends' opinions of my fashion choices are important to me.How people online view my fashion choices is important to me.CSR CSR communication(CSR1)α = 0.873 I only buy fashion garments from companies that are ethically or sustainably certified.I only buy fashion garments from companies that are transparent about their CSR practices (e.g. has a CSR report).I only buy fashion garments from companies that are consistent in their CSR practices.Personal values CSR(CSR2)α = 0.846 I only buy fashion garments from companies that consider consumer protection and employee welfare.I only buy fashion garments from companies that protect the environment.I only buy fashion garments from companies that contribute to local community development projects.Influencer influence (INFL) I follow various celebrities, bloggers and/or influencers online.The relationship I have with a social media influencer (e.g. a celebrity or blogger) informs my fashion choices.I am more likely to buy a product if an online influencer reviews it positively.I am more likely to like a brand if an online influencer reviews it positively.
*note that FCE yields no pathways for the revised model and has been excluded in this table.